BERT
Fewer than 5 people recommend it so far, so there's no written verdict. Here is what commenters said.
3recommend
1warn against
Nov 20, 2025first mention
Oct 24, 2026latest mention
Why people recommend it
“[BERT] -> [pooling] -> [probability calibrated SVM]”
“we used various bert variants to satisfy our usecases and still in uses”
Show 1 more
“it's still worthwhile to review these models and codebases”
Caveats and dissent
“don't screw around with fine-tuned BERT”
Often compared with
Threads where it came up
- The $25B Eigenvector – The Linear Algebra Behind Google (2006) [pdf] 1 mention
- Using LLM for confidence scoring is a bad idea 1 mention
- Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers 1 mention
- Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms 1 mention
- Kev: Tiny Jev-like family of decision models built on top of Qwen3.5 1 mention
- I built non-autoregressive decision models with RL a year ago 1 mention
Data: bert.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 13
“testing TextRank against BERT for 2-3 sentence article summaries”
“[BERT] -> [pooling] -> [probability calibrated SVM]”
“you can use BERT with sequence classification for simple classification tasks”
“we used various bert variants to satisfy our usecases and still in uses”
> And even if BERT wasn't woefully underintelligent for the task... have 100+ instances of BERT running locally faster than Jev API response times? Sweet rig you must have... why the heck do you need 100+ instances of bert. do you even attempt to research about this before? the laya paper show that you can do the similar stuff with jev using modern bert only: https://laya.convaiinnovations.com/ and even without the n
“We used to use BERT-based embeddings + semantic distance for classification / decision problems in new domains.”
“Something like BERT could be and has been tuned”
“it's still worthwhile to review these models and codebases”
“run this with a BERT model”
“uses a BERT based named-entity-recognition model running in your browser”
“You can actually generate surprisingly coherent text with minimal finetuning of BERT”
“don't screw around with fine-tuned BERT”
“using a trained BERT model as a reranker for Markov chain emmisions”